Companies are becoming increasingly global, which leads them to adapt their business to other markets. But, how can this adaptation be done without entailing a high cost? Many of them have opted to implement Artificial Intelligence solutions to adapt customer service in each context.  

This is where virtual assistants come into play, whose technology continues to grow and develop exponentially. In fact, customers prefer this type of attention when it comes to resolving their queries, ahead of a human agent. 

Virtual assistants are changing the structure of companies around the world. Not only improving the customer experience but also the internal management of companies. In this way, the simplest and most straightforward interactions are referred to virtual agents and the most complex queries to contact centers. 

Many international media outlets have reported on the great value of conversational agents. One of them is the Swiss magazine Fokus, which has compiled success stories of virtual assistants from all over the world. Among them is EME, Mutua Madrileña's conversational assistant, as an outstanding case in Spain. 

According to Gartner, by 2022, 70% of consumer interactions will involve emerging technologies such as the use of chatbots or conversational systems - based on Machine Learning. This would be a significant increase compared to the 15% of interactions that were recorded in 2018.

In this international context, the solution of providing multilingual support to users becomes important. This makes conversational assistants the technology of choice instead of chatbots. As chatbots do not offer a multi-language service.

Multilingual virtual assistants: when to deploy them

Before opting for a multilingual virtual assistant, companies need to ask themselves whether they need a conversational agent to speak more than one language. For this purpose, we distinguish between three groups of companies, which might fit this multilingual solution due to their context:  

• Companies operating in several countries. In this situation, a virtual assistant offers effective multi-language support without increasing costs. This solution can be a good strategy for companies that are planning to start operating in territories where different languages are spoken.  
• Organizations with more than one official language at their headquarters. Switzerland would be an example of this, where four official languages coexist. Other cases would be Spanish regions such as Catalonia, the Basque Country or Galicia.  
• International companies for internal communication. These systems often support the work of Human Resources. What if the company has offices in several countries? In this circumstance, it would be essential to offer multi-language support to your staff.

Once you have decided to go for a multilingual virtual assistant, the next step is to determine which technology is going to implement it. Thereby, we will look at the Dialogflow solution, as it is an advanced natural language model that is optimal for this type of situation. 

Multilingual agent in Dialogflow

Multilingual assistants in Dialogflow require the joint work of developers and linguists. Indeed, a conversational assistant is not just the result of technical work. In this type of project, personality and training are important details in its development.

Within the Dialogflow console, the multilingual function can be developed in several ways:

• An agent translated into several languages with Cloud Translation. It consists of creating an agent by selecting a primary language, such as Spanish, for example. Google Translate technology is used to be able to answer in several languages. This alternative has the advantage of having several agents with their respective independent languages. Therefore, the process would be to detect the user's language and, subsequently, send one agent or another depending on it. 
• A single agent with the option of several languages. It consists of creating a single agent with a primary language, which could be, in this case, Spanish. Thereafter, all the intents are duplicated for another secondary language, such as English. The task of translating the intents from Spanish to English would be the work of the linguistic profiles.
• Different agents for the same GCP project. Each agent has its language and is associated with the same Google Cloud Platform project. This option is only possible in Dialogflow CX, because in Dialogflow ES you cannot associate more than one agent to each GCP project. This is one of the differences between Dialogflow CX and Essentials. 

The latter two options are more accurate, but also more costly. In fact, their development requires the intervention of a computational linguistic team. However, in order to provide good customer service, the key is to adjust the language and personality to the culture and peculiarities of each country. 

Personality and training of a multi-language agent

The mission of professional linguists is not only to translate dialogues. Rather, their role is to incorporate cultural and idiomatic issues into the assistant. In other words, a conversational system has a defined personality that reflects the identity of the company. Part of their characteristics has a lot to do with the type of customer they are addressing.

This is where the computational linguistics team is responsible for adapting the agent's language and personality to the target environment, as well as adjusting to the culture, which differs even in regions under the same official language.

The company's personality

A company's personality is not rigid or immutable: it can vary according to the customers it addresses.

These changes, however slight, would be reflected in the conversational agent. Its character, for example, may vary according to the traditions and customs of the destination. Thus, the same virtual assistant can be more or less extroverted, serious or polite, friendly or distant... in order to adjust to the context of his or her interlocutor.

Another last aspect to take into account is the function performed by the assistant. 
If it is focused on getting leads, the agent redirects the user to fill in forms. But if you want to improve your lovemarks, you will change the tone of the conversation.

The country's culture

There are countries with the same official language, but with totally different cultures. This is the case of the United States and the United Kingdom, or Brazil and Portugal. In this case, professional linguists, trained in the philology of each country, have the task of training the virtual agent by modifying icons, words and expressions. 

Another influencing factor is the law. The regulations in each country determine what kind of responses the virtual assistant is allowed to make. For example, the exchange of certain personal data is not possible in certain countries. 

International virtual assistant

In conclusion, the increase in the use of virtual assistants goes hand in hand with the growth and globalization of companies, as well as with the need to provide a customer service that adapts to their language and cultural context. All of this leads us to think about offering a personalized quality service. 

In order to implement the internationalization of conversational agents, it is appropriate to have partners who are experts in the field. Computational linguists are responsible for the structuring and translation of languages. And, on the other hand, specialist developers are in charge of executing the entire IT part. 

Finally, for further information and in-depth knowledge about virtual assistants, you can go to the Dialogflow Experts website.